Fig. 10From: A simulation study on optimal scores for speaker recognitionPerformance loss with within-class variance mismatch between the training and deployment. The x-axis represents the distortion factor α on the within-class varianceBack to article page